11/20/24 - Episode 6 Data Dictionary
Text Summary
Eric Snyder: Welcome everyone to webinar number six. I'm Eric Snyder, I'm the executive director of the University of Rochester Medical Center's Wilmot Cancer Institute. Today, we'll be showing our Data Dictionary application. This is one of those things that everyone expects with any data-driven warehouse, but given its this team, of course, we had to think of a more innovative way to build it. Presenting today is Emily Strong, the Principal Developer of the Data Dictionary. Emily has been with us for a couple of years now, or a little over a couple of years, and in that time she's already done some pretty amazing stuff. I think the most striking thing to me about Emily is her ability to absorb any topic. I'm constantly amazed with every project Emily does. If it's not something she's done before, she'll come back the next day and absorb all the knowledge, and likely have a functioning prototype built, which I'm not even really exaggerating about. That's generally what happens, it's pretty impressive. So enough from me though, let's hand it off to Emily Strong, so without further ado.
Emily Strong: All right, thanks Eric for the introduction. Today I'll be talking about one of our newest applications, Data Dictionary, which is a surgical environment of healthcare data definitions. I'll first go through the motivation behind the development of Data Dictionary. I'll go through an overview of the application as well as its key features. I'll do a live demo of how to use the data dictionary and then talk about future work that we have in mind to continue to develop the app, as well as provide contact information for our team.
So, there were a few motivating factors behind the development of our Data Dictionary. With a vast data warehouse that's continually growing and changing, we needed a way to centralize all the information we have about these data. So, this would help our team effectively manage and utilize the data that we bring in. And by having all of this information in one place, we could encourage researchers, clinicians or other users to explore the data elements that may be available to them without needing constant support from our team. We also wanted to address data inconsistencies, which we commonly see when working with healthcare data. Oftentimes different groups or even the same group working on different projects might have multiple interpretations of a single concept. For example, a researcher working with some data might be talking about patient age but be referring to age at diagnosis. Whereas a different researcher may be even working with the same data might be talking about patient age, but actually be referring to patient age at first treatment. So, it's really important to provide a shared understanding or a standardized set of data definitions across healthcare groups to avoid any confusion that may arise when working with data and then finally, as I mentioned, our data warehouses continually changing. And so, we wanted a solution that would allow easily allow admins to update data information within some kind of application to promote the ongoing maintenance of our data definitions.
What is Data Dictionary:
So as I built data dictionary, I kept all of these motivating factors in mind. Our Data Dictionary, just as it sounds, serves as a centralized repository of information by containing definitions and metadata for over 300 tables. And all of this information is updated in real time, for example, if a table is added to our database, then our Data Dictionary automatically reflects this information. We also allow users to explore data elements through a search filter and tagging functionality. We address data inconsistencies by providing thousands of clear and concise data definitions to ensure accurate reporting for anyone who's using our data dictionary. And I created a special set of admin privileges that enable these users to edit table information, add and remove tags and add tables to the data dictionary right within the application.
Features:
So now I'll talk about key features of the Data Dictionary. When any user accesses the app, this is what they'll see on their screen, whether they are an admin user or not. So, at the top of the screen is a search bar where users can enter a free text search and optionally choose one of our currently defined filters to refine their results. Or instead of free text, they can select a bubble in the Hyperion app section to see tables linked to one of our team-built apps. Or at the bottom of the page, a user can choose a bubble in the data tag section to see tables that are linked to a particular phrase or term. So there's a few different options for users to make searches in our Data Dictionary.
When a user executes their search, this is an example of what a non admin user would see on the screen. So, the search and filter bar remains at the top of the screen to remind the user of what they've searched and also contains a statement with how many tables are in the results. So, for example, so the user knows whether they have to scroll through five tables or maybe like 100. Then in below the search bar, there is a block for each different table in the search results, and in that block is information about the table. So, at the top of the block is the table name, which in this case is BRFSS underscore NYS County. And the block also shows the number of rows and fields in that table, as well as a table description. And the number of rows and fields is updated dynamically or in real time and then there's also the tags and apps associated with that particular table. So, for this table, there's only one tag link to the table, which is on NYS public health record, and there are no Hyperion apps currently linked to it. And then below the rest of the table block is filled up with the names and brief descriptions of the fields in that particular table. So, this is the non admin view.
If we switch to the admin view, the page layout is generally the same, the search and filter bar stays at the top, but right below the search bar, there is an additional filter button added, where users can admin users can choose to view tables that are visible to non admin users, or they can choose to see tables that are currently in our database, but not yet have information in the data dictionary. The number of rows fields and table description stays the same, but then in the tag and app section, there is now an X on each tag or app block to allow admin users to remove a particular tag or app and there's also a plus button where users can choose to add a new tag, or if they want to add a new Hyperion app associated with this table. In the top right hand corner of each table block is a little eyeball icon that indicates whether this table is visible to non admin users, and this serves as a toggle. So, admin users can use that little icon to change visibility on or off for non admin users. And then throughout the table, the table block, there are several edit icons where admin users can choose to edit the table description or edit any of the field descriptions. Any changes that the admin users make are immediate, both on the front and back end. So, adding or removing tags or editing, for example, field description. These changes will show up on the screen, but also change field description, say, on the back end of the app.
So now I'll do a live demo of our Data Dictionary. So, this is the screen that shows up when you first access the data dictionary and first, I will show the non admin view. So, say I want to search interventions, but I only want to look through, I only want my results to show to show tables where interventions is in the table name. So, I would choose the filter table name, and then click search. So, I can see there's two tables that show up in my results. The first one is CT underscore go underscore arm group interventions. So, if I click on the table name, then some information will expand about that table. I can see the number of rows and fields as well as description in that table, and then the tags and apps link to that table. So there's one tag currently linked to the CT Gov Arm Group Interventions Table, which is Clinical Trials.Gov. And there are currently three Hyperion apps linked to this table, which are Public Trial Viewer, via and Trial Tracker. So, I can see all the field names and descriptions for this table and then scroll down to see other tables in my results. So, this once again is the non admin view.
If I want to look at the admin view, the front page is exactly the same. But say this time I want to search, I'm going to click on a data tag to see the tables linked to a particular term and I'm going to click on NYS public health record and I don't have to click search or anything that automatically directs to the search results, and the search bar automatically fills in as well. So even though I didn't enter anything in the search bar, and I clicked on that data tag, the search bar fills in. So, I still know what I search for. So, I can see the additional filter button. Now that I have the Admin View. I can expand this table information and now the access for removal and pluses for addition show up in the tag and app section. So, if I wanted to add a tag to this table, then I could click the plus sign and then I could add a tag called public health to this table and click add tag. So, this change once again is immediate both on the front and back end and then if I want to remove this public health tag that I just created, then I would click on the X and I get a prompt that ask me, are you sure you want to remove this tag? And then I just click remove. and now we can see that tag is no longer listed under this table.
Similarly with apps, I can click the plus, and I have a drop-down menu with names of all our Hyperion apps to choose from. So, say I want to add Cohort Builder to this be linked to this table, then I would click add and now I can see there’s one app linked to this table. If want to remove that, the similar prompt comes up and I can just click remove. I can also edit table descriptions and field descriptions. Say I want to edit one of these filed descriptions for the County field. Currently this description is NYS County. I can click the edit icon in a form comes up for to enter a new field description or edit the existing one. If I just put on NYS County name to edit that field description and make it more specific, then I can click update and now we see there’s a new field description there. And finally up in the right hand corner of this table block, there that little eyeball icon. Currently this table is visible to non admin users. If I click that eyeball, then there's now a slash through it, which changes the visibility on the front and back end. So now admin users can no longer see this table in their search results. So, I'll just undo that. And then I've turned the visibility back on. So this is the, or that was the live demo, and now I will go back to our, or my slides.
Future Enhancements:
For some future work for the Data Dictionary, we as I mentioned our Data Warehouse is continually changing and so we need to add and revise any information that we bring in to provide accurate and to up to date health care data definitions for any data dictionary users. And for phase two of the project, we’d also like to implement baseline validated queries, which would allow users to quickly execute commonly used data queries. For example, If I wanted to get all patients that have had chemo, but I might choose a query that would give me these patients from a list and then this ensure that each time that I want to grab these patients, then I’m reproducing that set of patients. So it's promoting data reproducibility, as well as eliminating the time necessary to, you know, write out the whole query and find what tables I need.
I was responsible for the development of this application, as Eric mentioned, the back end front end development design, writing all the Data Dictionary definitions up until this point, as well as testing and documentation.
If you would like to learn more about our team, The Technology and Innovation Group, then you can scan the QR code on this page around the screen or enter the link on the screen in your search bar to see all the cool projects that we're working on, read more about our team members and watch our previous webinars. So that's all I have for this webinar and now I'll open it up for questions.
Eric Snyder: Thanks Emily. It's fantastic. Before we had any questions I wanted to thank everyone for joining.
Q/A Session:
Q: So it looks like it would be interesting to see patterns of utilization. Do you currently log what people are searching for?
A: Yes, we do. And this is for security purposes as well as if we wanted to analyze, you know, any search patterns in the future.
Eric Snyder: So it looks like there's any more so if there isn't any more, then we will close it here and I think everyone for joining.
Thank you.